Claude Opus 5 (Adaptive Reasoning, Max Effort): benchmark results

Provider: Anthropic. Released 2026-07-24. Access: API.

Unified ELO 1744 ± 1, rank #38 of 1761 rated models, from 25 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Artificial Analysis Intelligence Index54.05Intelligence Index99.5
AA Humanity's Last Exam54.87Accuracy (%)99.3
AA GDPval1739.24ELO99.2
AA Omniscience - Business53Accuracy (%)99.2
AA Omniscience - Science, Engineering & Mathematics57.52Accuracy (%)99
AA Omniscience - Humanities & Social Sciences58.6Accuracy (%)98.4
AA-Omniscience Accuracy60.87Accuracy (%)98.4
AA CritPt29.14Accuracy (%)98.1
AA Omniscience37.07Score98.1
AA Omniscience - Software Engineering (SWE)86Accuracy (%)97.9
AA Omniscience - Health52.27Accuracy (%)97.7
Riemann-bench68Score (%)97.3

Interactive version: theaggregate.ai/model?slug=claude-opus-5-adaptive-reasoning-max-effort · How It Works · Data refreshed daily, snapshot 2026-09-05.